Upregulation of adenosine A2A receptor in astrocytes is sufficient to trigger hippocampal multicellular dysfunctions and memory deficits

Abstract

Adenosine is an ubiquitous neuromodulator that ensures cerebral homeostasis. It exerts numerous functions through the activation of G-protein-coupled adenosine receptors (ARs), in particular A₁ (A₁R) and A_2A (A_2AR) receptors. Interestingly, A_2AR levels are upregulated in cortical and hippocampal regions in several pathological conditions such as Alzheimer's disease, tauopathies or epilepsia. Such abnormal upregulations have been particularly reported in astrocytes, glial cells that play a key role in regulating synaptic plasticity. However, the overall impact and the underlying mechanisms associated with increased A_2AR in astrocytes remain poorly understood. In the present study, we induced the upregulation of A_2AR in hippocampal astrocytes using dedicated AAVs and comprehensively evaluated the functional consequences in 4 months-old C57Bl6/J mice. Our results show that A_2AR upregulation primarily promotes alterations of astrocyte reactivity, morphology and transcriptome, with a link to aging-like phenotype as well as secondary impairments of neuronal excitability and microglial phenotype. These changes driven by a restricted A_2AR upregulation in hippocampal astrocytes were sufficient to induce impairments of short-term spatial memory and spatial learning. This study highlights the impact of astrocytic A_2AR upregulation, as seen in various neurological conditions, on the development of a detrimental multicellular response associated with memory alterations and provides an additional proof-of-concept for the value of targeting this receptor in different neurodegenerative conditions.

Fig. 1. AAV-based astrocytic A_2AR upregulation in the CA1 hippocampus in mice.

A Astrocytic A_2AR upregulation (or the GFP as control) has been obtained following bilateral hippocampal injection (CA1) of AAV2/9 carrying a transgenic murine A_2AR (AAV-A2A; blue) or enhanced form of GFP (AAV-GFP; white) genes under the control of the GfaABC1D astrocytic promoter. B Representative images of A_2AR immunostaining highlighting A_2AR upregulation in the mouse CA1 hippocampus (scale bar=500 μm, upper picture) and, at higher magnification, the distribution within the stratum oriens (SO), pyramidal layer (PL) and stratum radiatum (SR) sublayers of CA1 (scale bar=60 μm, lower picture). C Representative images of A_2AR co-immunostainings with the astrocytic markers GFAP (a), Sox2 (b) and S100β (c), neuronal marker NeuN (d) and microglial marker Iba1 (e) highlighting the exclusive expression of A_2AR in CA1 astrocytes (scale bar=20 μm).

Fig. 2. A_2AR upregulation in hippocampal astrocytes impacts their reactivity and complexity.

A Representative images of GFAP immunostainings in the CA1 of AAV-GFP (left panel) and AAV-A2A (right panel) animals (scale bar=100 μm; magnification in red inserts scale bar=50 μm). B Quantification indicates a significant increase of the GFAP⁺ staining in the hippocampal CA1 astrocytes of AAV-A2A mice as compared to AAV-GFP controls (N = 5-6 mice/group; 3-5 sections analyzed/animal; **P < 0.01 vs. AAV-GFP; Mann-Whitney test). C Representative images of GFAP (green), STAT3 (red) and DAPI (blue) co-immunostainings in the hippocampal CA1 area of AAV-GFP (upper panel) and AAV-A2A (lower panel) animals (scale bar=50 μm). D A significant increase of the STAT3⁺ staining in the GFAP⁺ astrocytes of AAV-A2A mice was observed as compared to AAV-GFP controls in the hippocampal CA1 area (N = 4-6 mice/group; 2-3 sections analyzed/animal; **P < 0.01 vs. AAV-GFP; Student’s t-test). E, F A retro-orbital injection of an AAV-PHP.eB carrying tdTomato gene under the control of the GfaABC1D astrocytic promoter in both AAV-GFP and AAV-A2A animals was performed (E) in order to express tdTomato protein in sparse CA1 astrocytes (cytosol and arborization; scale bar=60 μm; F). G Isolated Tomato⁺ astrocytes of the stratum radiatum CA1 sublayer were imaged by high-resolution imaging (a) and 3D-reconstructed using Imaris software (b). H–J An increase in the intersection number, as shown by Sholl analysis (***P < 0.0001 vs. AAV-GFP, Two-Way ANOVA; H) as well as of the number of the astrocytic processes (N = 14-17 astrocytes from N = 5-6 mice/group; **P < 0.01 vs. AAV-GFP, Student’s t-test; I) was found in the stratum radiatum CA1 astrocytes of AAV-A2A condition as compared to the AAV-GFP controls. That was accompanied by a reduced process length (N = 14-17 astrocytes from N = 5-6 mice/group; *P < 0.05 vs. AAV-GFP; Student’s t-test; J) (K–L) No change in the overall astrocyte volume (P = 0.40; K) or the astrocytic soma volume (P = 0.74; L) could be observed AAV-A2A animals as compared to the AAV-GFP controls (N = 14-17 astrocytes from N = 5-6 mice/group). Values are represented as mean ± SEM. Stratum oriens (SO); pyramidal layer (PL); stratum radiatum (SR) sublayers.

Fig. 3. Astrocytic A_2AR upregulation in hippocampus is associated with an aging-like astrocyte signature.

A The principal component analysis (PCA) performed on RNA-seq datasets shows that the two conditions AAV-GFP (white dots) and AAV-A2A (blue dots) are nicely separated on the first Principal component (PC1). B Volcano plot representing the deregulation of 1127 genes in the astrocytes of AAV-A2A vs. AAV-GFP mice using cut offs of |Log2 fold-change (FC)|>0.32 and adjusted P-value (Padj)<0.05. 916 genes were found downregulated and 211 upregulated, including the Adora2a gene. C Top 10 genes significantly downregulated (red) (i) and upregulated (green) (ii) in hippocampal astrocytes of AAV-A2A animals vs. AAV-GFP controls with |Log2 fold-change (FC)|>0.32 and Padj<0.05. D, E Functional enrichment analysis of GO Biological Process annotations using DAVID (upper panels) and GSEA analysis (lower panels) associated to (D) downregulated genes and (E) upregulated genes. F Representative images of GFAP (green), YAP (red) and DAPI (blue) co-immunostainings in the CA1 of AAV-GFP (upper panel) and AAV-A2A (lower panel) groups (scale bar=50 μm). Inserts represent YAP⁺ nuclei (dashed lines) in both groups (scale bar=10 μm). G A significant decrease in the percentage of YAP⁺ cells among GFAP⁺ astrocytes as well as a reduction of the YAP staining within the nuclei of GFAP⁺ astrocytes were observed in the CA1 hippocampal area of AAV-A2A mice vs. AAV-GFP controls (N = 4-5 mice/group; 3 sections analyzed/animal; *P < 0.05, **P < 0.01 vs. AAV-GFP, Student’s t-test). H Representative images of S100β (green), HMGB1 (red) and DAPI (blue) co-immunostainings in the CA1 of AAV-GFP (upper panel) and AAV-A2A (lower panel) groups (scale bar=50 μm). Inserts represent HMGB1 staining (red lines) in S100β⁺ cells (green lines) in both groups (scale bar=7 μm). I A significant decrease of the HMGB1 staining intensity in S100β⁺ astrocytes was observed in the CA1 hippocampal area of AAV-A2A mice vs. AAV-GFP controls (N = 4-5 mice/group; 3-4 sections analyzed/animal; **P < 0.01 vs. AAV-GFP, Student’s t-test).

Fig. 4. Astrocytic A_2AR upregulation in hippocampus alters neuronal activation.

A Phosphorylation of the GluN2B subunit (NMDAR) at Y1472 and of the GluA1 subunit (AMPAR) at pS831 in synaptosomal fraction of AAV-GFP and AAV-A2A hippocampi. Quantification shows a significant increase of pS831 GluA1 in AAV-A2A vs. AAV-GFP mice (*P < 0.05; **P < 0.01 vs. AAV-GFP; Student’s t-test; N = 6/group). B Bilateral injection of an AAV5 carrying the hM3Dq gene under the control of the CAMKII neuronal promoter allows the activatory Gq-coupled DREADD-mCherry receptor expression in hippocampal neurons of AAV-GFP and AAV-A2A animals. Two months following the AAV co-injections, an i.p administration of the exogeneous and synthetic hM3Dq ligand (clozapine-N-oxyde, CNO; 5 mg/kg), or saline (NaCl; 0.9%) as a control, induced neuronal activation measured by the immediate early genes (IEGs) expression. C Representative images of A_2AR (green), RFP-amplified hM3Dq-mCherry (red) and DAPI (blue) co-immunostainings in CA1 hippocampal area (scale bar=200 μm; magnification: scale bar=10 μm). D–F The mRNA expression of Dusp1 (D), JunB (E) and c-fos (F) was found significantly higher in CNO-treated vs. saline-treated animals. Notably, IEGs expressions were significantly enhanced in AAV-A2A as compared to AAV-GFP mice in CNO-treated conditions. (N = 4-5 mice/group; ^##P < 0.01; ^###P < 0.001: CNO-treated vs. respective saline-treated animals; ⁺P < 0.05; ⁺⁺⁺P < 0.001 in AAV-A2A + CNO-treated animals vs. AAV-GFP ⁺ CNO-treated; One-Way ANOVA followed by Tukey’s post-hoc test). G Representative images of c-Fos immunostaining in the CA1, CA2 and DG regions of hippocampus in AAV-GFP (upper panel) and AAV-A2A (lower panel) animals treated with CNO (scale bar=400 μm). H As expected, the intensity of c-Fos was significantly higher in CNO-treated vs. saline-treated animals. c-Fos immunostaining intensity was found significantly higher in AAV-A2A + CNO as compared to AAV-GFP + CNO controls. (N = 4-5 mice/group; 2 sections analyzed/animal; ^##P < 0.01; ^###P < 0.001: CNO-treated vs. respective saline-treated animals; ⁺P < 0.05 in AAV-A2A + CNO-treated animals vs. AAV-GFP + CNO-treated; One-Way ANOVA followed by Tukey’s post-hoc test). I Recordings of local field potentials showing the bursting activity of the CA1 region of hippocampal slices of AAV-GFP (upper trace) and AAV-A2A (lower trace) mice during application of 4-AP/Low Mg²⁺ -containing extracellular solution. Insets (red dotted rectangles) show a magnification of a single burst in both groups. J The burst frequency was significantly higher in AAV-A2A animals as compared to the AAV-GFP controls (N = 8-9 slices/group from 3-4 mice/group; *P < 0.05 vs. AAV-GFP using Mann-Whitney test). Values are represented as mean ± SEM.

Fig. 5. Astrocytic A_2AR upregulation in hippocampus alters microglial phenotype.

A Representative images of Iba1 immunostaining in AAV-GFP (left panel) and AAV-A2A (right panel) conditions in the CA1 hippocampal area (scale bar=100 μm; magnifications: scale bar=10 μm). B, C A significant increase of the Iba1⁺ staining was observed in the hippocampal astrocytes of AAV-A2A condition as compared to the AAV-GFP control (N = 4-6 mice/group; 3-5 sections analyzed/animal; *P < 0.05 vs. AAV-GFP; Student’s t-test; B) with no change in the density of Iba1⁺ cells (C). D Representative images of Iba1 immunostaining (a) and 3D reconstruction (b,c) (scale bar=15 μm; magnification: scale bar=2 μm). E Decrease in the intersection number (complexity), as shown by Sholl analysis, was found in the microglia of AAV-A2A mice as compared to AAV-GFP controls (N = 10-15 microglia from N = 5 mice/group; Two-Way ANOVA). F Volcano plot representing the deregulation of 471 genes (DEG) in the microglia of AAV-A2A vs. AAV-GFP mice using cut offs of |Log2 fold-change (FC) | > 0.32 and adjusted P-value (Padj)<0.05. 214 genes were found downregulated and 257 upregulated. G, H Functional enrichment analysis of GO Biological Process annotations using DAVID (upper panels) and GSEA analysis (lower panels) associated to (G) downregulated genes and (H) upregulated genes. I Representative images of Iba1 (red), CD68 (yellow) and DAPI (blue) co-immunostainings in the CA1 hippocampal area in AAV-GFP (higher panel) and AAV-A2A (lower panel) groups (scale bar=10μm; magnification: scale bar=2 μm). J A trend for a significant reduction of the CD68⁺ staining in the Iba1⁺ cells of the stratum radiatum CA1 sublayer was found in AAV-A2A as compared with AAV-GFP controls (N = 5-6 mice/group; 3-4 sections analyzed/animal; P = 0.08 vs. AAV-GFP; Mann-Whitney test). Values are represented as mean ± SEM.

Fig. 6. Astrocytic A_2AR upregulation in hippocampus impairs short-term spatial memory and spatial learning.

A Schematic representation of the Y-Maze experimental paradigm described in the Material and Methods section. B The discrimination index, taken as a measure of the short-term spatial preference for the novel arm in the Y-maze task, was significantly lower for the AAV-A2A animals as compared with AAV-GFP controls (N = 9 mice/group; *P < 0.05 vs. AAV-GFP; Student’s t-test). C Schematic representation of the Barnes Maze task experimental paradigm described in the Material and Methods section. D–F During the learning phase of the Barnes task, the distance (D), the primary latency (E) and the number of primary errors (F) were found significantly higher for the AAV-A2A animals as compared to AAV-GFP controls (N = 11-12 mice/group; ***P < 0.001; Two-Way ANOVA). G No change regarding the time spent in the target quadrant during the retention phase was found between AAV-A2A and AAV-GFP animals (N = 11-12 mice/group; P > 0.05; Student’s t-test). Values are represented as mean ± SEM.